The next step in data analytics: Getting from historical to actionable analytics

There is often a discernable difference between the current stage of a hospital's analytics maturity and where the organization desires to be on the maturity curve.

This content is sponsored by RelayHealth.  

While the concept of analytics is not new to healthcare, the scope has been rapidly changing over the past few years. Most U.S. hospital systems have begun to transform from standard historical reporting systems focused on purely inpatient data to experimenting with predictive and population-oriented systems incorporating data well beyond the four walls of the hospital.

Advancing technologies have created new opportunities for leveraging data to improve clinical decision-making, reduce readmissions and identify places to improve efficiency and trim costs. For hospital systems that can capture, store, clean and manipulate data well, the payoffs are substantial: healthier patients, higher consumer and physician satisfaction rates, greater operating margins and better reimbursement rates.

The advantages of this next generation data analytics are clear. Yet, most systems today still struggle to attain this level of sophistication. For one, most organizations don't possess a team of in-house data scientists experienced in mining data from disparate sources, transforming that data into insights and deploying those insights to drive business optimization and clinical action. While evolving across most organizations, the role of "analyst" is still a very loose and widely varied one focused more on the historical paradigm of analytics or reporting than the new frontier paradigm where analytics proactively informs a fully empowered multistakeholder data-driven culture. Data is seen as the byproduct of multiple IT systems and a barrier instead of an enabler due to the lack of easy, repeatable access to high reliability information.

Just over 50 percent of health data scientists said the most significant barrier to analytics is not knowing what data to collect or how much of it, followed by a lack of organizational clarity on what to do with data and what to look for when analyzing it, according to a 2015 Stoltenberg Consulting survey

"People are frustrated the analytics of yesterday are no longer adequate and the analytics needed for transformation today are frustrating and challenging to get to. A purposeful, fully resourced data strategy is the key pivot point to get from counting and trending to predicting and preventing an ultimately to a learning system that is prescribing evidence driven protocols at the point of care," Tina Foster, vice president of business advisor services for RelayHealth, told Becker's Hospital Review

To develop a successful analytics strategy, Ms. Foster says it is always good to go back to the basics of solving for an existing business need. "Organizations have to define value for themselves — whether it's reducing Medicare readmissions or cutting costs in a specific procedure — and then identify key metrics of success, key sources of reliable data for the numerator and denominator, key data science models to get to meaningful information and use the data in a focused way to achieve that business imperative," Ms. Foster says.

This article examines challenges associated with an incomplete data strategy as well as the advantages of using small, business-oriented analytics projects to drive data maturity and scale.

Lack of strategic clarity is costly

When administrators and clinicians dive into data projects without a defined business need, hospital systems can experience a number of costly setbacks, including wasting time and resources on tangential data re-work or making decisions based on incomplete data.

Some administrators may associate data analytics maturity with advanced intelligence platforms. Already 65 percent of health IT professionals reported their organization had a business intelligence or analytics solution in place, according to a March 2017 poll by HealthITAnalytics.

Technology is important to sort through sizeable amounts of patient-generated health data. However, technology is only as valuable as the insights it yields. Feeding incomplete or low-quality data into an advanced analytics platform will produce inaccurate or misleading results. When this happens not only are the outcomes not optimized, but it typically results in low clinician confidence which in turn creates more skepticism around the data and makes subsequent transformation even more difficult.

"People are highly skeptical of the data in healthcare," Ms. Foster says. "It's diverse, scattered and cumbersome. Health systems spend a bunch of time doing manual reconciliation — by the time you get to analyzing the data, the reliability isn't very high."

As is, data scientists and many health systems spend a substantial amount of time making data usable. The granular nature of the job — what data scientists call "data wrangling" or "data janitor work" — requires substantial labor and cost. Data scientists reported spending up to 80 percent of their workday collecting and preparing unruly, disparate healthcare data, according a survey conducted by The New York Times. That leaves only 20 percent of the day to actually use and analyze data to drive insights that bring positive change to the organization.

Instead of focusing on technology and data wrangling, Ms. Foster recommends hospital systems looking to improve their data analytics begin by establishing a very clear, specific business goal that can be achieved using data.

Advantages of starting small and scaling up

Ms. Foster identified three key advantages hospitals gain by starting with small, focused analytics projects.

1. When projects start small, stakeholders have a clear understanding of their role. Articulating clear business or operational needs enables people to speak about analytics and data in terms they can visualize and understand. For example, a business objective might be reducing readmission within 30 days of discharge for Medicare patients admitted with acute myocardial infarction. Putting a data project in plain terms makes it more accessible and actionable for team members across disciplines.  

Setting business needs first also lays the groundwork for what functions analytics need to support. This, in turn, enables the analytics team to determine appropriate data sources and set parameters for data quality, allowing data scientists to focus on only cleaning the data relevant and useful for a reliable, analysis-ready database.

When business and analytics needs are clearly established, data quality can also be better established: timeliness, standardization, accuracy and completeness of data make the most sense when elements of data quality are framed by business and analytics contexts. Then, staff can begin the hard work of creating and maintaining a reliable database. 

Consider the above Medicare example. The business objective requires building a predictive model that can stratify the patient population according to certain risk factors to determine those patients with the greatest risk of readmission. This gives data scientists a clear understanding of what patient information to look for — age, zip code, socioeconomic status —where to find it and how to format it.

2. Accomplishing a small business goal builds morale and support. In a well-defined space, analytics can make products and clinical programs work better and faster at a lower cost — and getting an early win is critical to gaining stakeholder support.

"As you begin to demonstrate results [using data analytics] — revenue, cost savings, better health outcomes — you gain confidence, building a groundswell of consensus and support to take on more programs, then even bigger programs," Ms. Foster says. "Then, you can begin to focus on a data strategy that allows you do the same thing on a much greater scale."

3. Tangible results from analytics programs can be used to fund additional projects. Seeing concrete results of analytics programs is important to sustain employee motivation. Tangible results can include cost savings, reduced overtime hours, additional revenue or higher patient satisfaction. Whatever the benefits, Ms. Foster says hospital administrators can use the fruits of their analytics program to fund the next analytics initiative, thereby making the "one process improvement at a time" approach to analytics self-sustaining.

"Start with one project, do it well and get a win," Ms. Foster says. "Then you can take the yield — whether that be revenue enhancement, cost savings or quality improvement — and use it to pay for the next analytics initiative."

A call to action: Assess your analytics capability

Moving up the analytics maturity model is no easy feat. That's why some hospitals see value in partnering with data strategy and business advisors for guidance in developing data competencies. 

"As organizations begin to address data challenges and aspire to more complex use cases, they need someone who can help them avoid the pitfalls and do the heavy lifting," Ms. Foster says. "Professional services can provide that expert knowledge and experience from doing [data analytics] every day."

Advisor firms like RelayHealth have years of experience in assessing the strengths and challenges of health systems' analytics, as well as in improving the personnel, tools and processes in data management.

"Data is the most strategic asset an organization has, and yet most people treat it like a byproduct of their EMR or financial system," Ms. Foster says. "Health systems that invest in their data and embrace their data are on the path to long-term viability."

More articles on data analytics and precision medicine: 

23andMe to expand with $200M in latest funding round
UC Davis to use precision medicine in pancreatic cancer care
Global genomic data platform WuXi NextCODE raises $240M

© Copyright ASC COMMUNICATIONS 2017. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months